Carbothermal Reduction of Boron-bearing Iron Concentrate and Melting Separation of the Reduced Pellet

Abstract

Comprehensive utilization of low grade ludwigite is critical important for the safe supply of boron resource of China. Boron-bearing iron concentrate is the intermediate product of the ore dressing process of crude low grade ludwigite and extracting boron from it has great effect on the total boron yield. Pre-reduction of the boron-bearing iron concentrate is one of the important steps of the pyrometallurgical boron and iron separation process. The carbothermal reduction study of boron-bearing iron concentrate was carried out in the present work. The reduction rate was improved with the increasing of heating temperature and carbon content. The appropriate temperature and C/O (mole ratio of fixed carbon to reducible oxygen) were 1200 to 1300°C and 0.8 to 1.2, respectively. In addition, the microstructure and phase evolution of the pellet during the reduction process were characterized by means of SEM and XRD. The apparent activation energy of the iron oxide reduction in the composite pellet was 114.32 kJ/mol based on the first-order reaction model. The iron and slag separate well when the reduced pellets were heated at 1550°C. The B₂O₃ content of the slag and the boron element content of the pig iron were 10.8 wt% and 0.74 wt%, respectively. The main minerals in the slag were olivine, kotoite, periclase and spinel after slow cooling, and the efficiency of extraction of boron (EEB) of the slag was 68.4%. The boron-rich slag and boron-bearing pig iron can be used as the raw materials for boron extraction and steel making.

1. Introduction

Boron compounds, chiefly borates, are commercially important and there are more than 300 end uses for borates. World consumption of borates was projected to reach 2.0 million metric tons of B₂O₃ by 2014, compared with 1.5 million metric tons of B₂O₃ in 2010. China makes an important contribution to the increasing of the total boron consumption in the world due to its large demand.¹⁾ China has low grade boron resource. Traditional boron resource-szaibelyite ore of China is going to run out and its grade becomes much lower than before. It is estimated that China will import about 1.5–2 million tons of boron minerals in 2020. Fortunately, there is 0.28 billion tons of low grade ludwigite deposite in Liaoning and Jilin province, which accounts for 57.88% boron reserve and 1% iron reserve in China, is the main alternate resource of conventional szaibelyite ore. It can’t be directly utilized as any single kind of ore by traditional methods because of their low grades. The average boron and iron content of the ore is about 7.23 wt% and 30.65 wt%, respectively.^2,3)

Carbon composite pellet uses the mixture of fine iron ore with fine carbonaceous materials as the raw materials and is agglomerated by rotation balling and high pressure pressing. It has been known that the reduction rate is very high because of the direct contact of the iron oxide with carbon. The iron production time can become much shorter than the conventional iron making process. In recent years, increasing attention has been paid to the carbon composite pellet by many strategists and researchers in iron making industry. The iron ore-coal mixture is used in the direct reduced iron (DRI) process, such as Inmetco, Fastmet and Comet.^4,5,6) In order to obtain high quality slag free hot metal or DRI, several new processes, such as Fastmelt, Hi-QIP and ITmk3 process, have also been proposed.^7,8,9) The above technologies can be used to process traditional iron ore, complex iron ore and steel wastes. These processes have relatively simple and less expensive facilities and produces high quality iron, and they are also of great significance to environmental protection and resource saving.

At present, boron-bearing iron concentrate and boron concentrate can be effectively obtained by ore dressing method from the crude iron ore. The boron concentrate is good raw material for boron industry after activation roasting and the extraction of boron from boron-bearing iron concentrate is the major task in order to improve the total yield of boron. The boron and iron separation of boron-bearing iron concentrate can be realized through selective reduction and melting method. The separated boron-rich slag is used as the raw materials for borax production. This is an appropriate flow sheet for the comprehensive utilization of low grade ludwigite.^10,11,12) In the present work, a new coal-based reduction-melting process using carbon composite pellet for iron and boron separation from boron-bearing iron concentrate has been proposed by the authors’ team. The proposed flow sheet is shown in Fig. 1. The experiments of carbothermal reduction of boron-bearing iron concentrate and high temperature melting separation of reduced pellet were performed at laboratory scale.

Fig. 1.

Schematic flow sheet of the utilization of boron-bearing iron concentrate.

2. Experimental

2.1. Raw Materials

The boron-bearing iron concentrate used in this study was obtained from Liaoning Province, China. The chemical composition of the complex ore sample was shown in Table 1. The particle size of the concentrate was 90 wt% below 200 mesh. The mineralogical analysis of the concentrate was investigated by XRD analysis and the result was listed in Fig. 2. The result indicated that the main mineral phases were magnetite (Fe₃O₄), chrysotile (Mg₃[Si₂O₅](OH)₄) and ludwigite ((Mg,Fe)₂FeBO₅). The sulfur content of the ore was as high as 0.54 wt%. The microstructure of the boron-bearing iron concentrate was shown in Fig. 3.

Table 1. Chemical composition of boron-bearing iron concentrate (wt%).

TFe	FeO	B2O3	MgO	SiO2	Al2O3	CaO	P	S
48.3	22.9	5.26	15.3	8.09	0.32	0.42	0.027	0.54

Fig. 2.

XRD analysis of boron-bearing iron concentrate.

Fig. 3.

SEM-EDS analysis of the boron-bearing iron concentrate.

Pulverized anthracite coal was used as the reducing agent, which was screened by 1 mm opening sieve after ball-milling. The chemical composition of the coal was listed in Table 2, which revealed that the coal was high in fixed carbon, and comparatively low in ash, volatile, sulfur and phosphorus. It was a good reducing agent for the direct reduction process. The reactiveness of the pulverized coal was assessed by chemical reaction ability between pulverized coal and carbon dioxide (30 mL/min) through thermogravimetric method with a heating speed of 10°C/min. The result was shown in Fig. 4 and the maximum reaction rate occurred at about 1150°C.

Table 2. Chemical analysis of the reducing agent (wt%).

Proximate analysis				Ash analysis					S	P
FCd	Vd	Ad	Mad	SiO2	Al2O3	Fe2O3	CaO	MgO
81.40	6.40	11.10	1.10	39.95	32.04	12.66	4.14	0.84	0.34	0.032

M_ad: Moisture (dry basis), A_d: Ash (dry basis), V_d: Volatile Matter (dry basis), FC_d: Fixed Carbon (dry basis).

Fig. 4.

Reactiveness thermogravimetric curve of the pulverized coal.

2.2. Experimental Procedure

The experiments included isothermal reduction and high temperature melting separation two set of tests.

In the isothermal reduction test, the boron-bearing iron concentrate and pulverized coal were mixed together with mole ratio of “fixed carbon” in the coal to “reducible oxygen” in the iron oxide (C/O) of 0.8, 1.2 and 1.6. The ore/coal mixture was completely mixed by rigorously stirring for about 30 min. The prepared mixture was pelletized in a horizontal twin roller machine. The size of the pillow shape pellet was 40×30×20 mm and was dried at 120°C for 10 h. Reduction process was performed in a MoSi₂ resistance furnace, under a high purity N₂ flow of 5 L/min. In the experiment, the dry green composite pellets were heated at the temperatures of 1100, 1200, 1300 and 1350°C for certain time. Once the reduction experiments finished, the samples were taken out of the furnace and cooled to ambient temperature under the protection of nitrogen. The metallization degrees (η=(MFe (mass content of metallic iron, wt%)/TFe (mass content of total iron, wt%))×100%) of the reduced pellets were examined by chemical analysis. Some of the reduced pellets were cut, mounted in epoxy resin, polished and examined by electron microscope with an energy dispersive spectroscope (SEM-EDS) to interpret the reduction mechanism. The phase evaluation during the reduction process was characterized by powdery X-ray diffractometer (XRD).

For the melting separation test, pellets with a C/O of 1.2 were first reduced at 1300°C for 15 min, and then were cooled under the protection of nitrogen. The reduced pellets were crushed to below 1 mm and the powder sample was loaded in a graphite crucible with weight of 100 g. The sample was heated at high temperatures in the same furnace to realize the iron and slag separation to enrich the boron into the slag phase. After melting separation, the furnace was shut down and the sample was cooled to room temperature. The chemical compositions of the separated iron and boron-rich slag were examined by chemical analysis. The microstructure and the efficiency of extraction of boron (EEB) of the boron-rich slag were evaluated by SEM-EDS, XRD and the normal pressure alkaline leaching method.¹²⁾ The EEB was defined by Eq. (1).   

$$\begin{array}{l}\text{EEB}=[(\text{Boron amount of primary slag}\\ \hspace{1em}\hspace{1em}\hspace{0.17em}-\text{Boron amount of residual slag)}\\ \hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{0.17em}\text{/Boron amount of primary slag}]\times 100\% \end{array}$$(1)

It was an index to reflect the boron extraction rate of the boron mineral by the CO₂-soda borax production process.

3. Experimental Results and Analysis

3.1. Isothermal Reduction Behaviors

Effect of temperature on the reduction behaviors of the boron-bearing iron concentrate/coal composite pellets is shown in Fig. 5. It can be observed that the reduction rate is highly dependent on the temperature and is improved by temperature increasing from 1100 to 1350°C. The temperature affects greatly at low temperature range from 1100 to 1200°C, and smaller when increasing from 1200 to 1350°C. Traditionally, the time for the rotary hearth furnace (RHF) turning a round is a little more than 20 min. At final reduction point in the experiment, the metallization degree of the pellets reduced at 1200 to 1350°C are all more than 85% and the metallization degree of the pellet reduced at 1100°C is only about 60%. Thus the reduction temperature will be better if it is above 1200°C.

Fig. 5.

Variation of metallization degree of coal char composite pellets over time at different temperatures.

It is well known the reduction of iron oxide in the carbon composite pellet proceeds mainly through the gaseous intermediates CO and CO₂. The solid carbon transfers into gaseous reducing agent CO through the Boudouard reaction (C+CO₂=2CO), which greatly depends on temperature and plays an important role in the reduction process.^13,14) The Boudouard reaction proceeds at viable rate when the temperature is higher than 1150°C according to the result shown in Fig. 4 and the reduction rate of the pellet will be at a faster speed when heating temperature is more than 1150°C. The reactiveness of the pulverized coal agrees well with the present reduction results.

The effect of carbon content (i.e. C/O) on the metallization degree of the pellets is studied at 1300°C. The values of C/O vary from 0.8 to 1.6. The reduction curves are presented in Fig. 6. It shows that the metallization degree increases with the increasing of C/O to some extent. The pellet with C/O of 0.8 is short of reductant at the later period of the reduction stage and the final metallization degree is less than 80%. As a matter of fact, when the content of pulverized coal increases, the Boudouard reaction will proceed at a faster rate to provide more CO and the reduction is then improved. The carbon content will be more appropriate if the C/O is a little lower than 1.2. It also can be seen that the final metallization degrees of all the reduced samples can’t reach 100% and the reason for this is very complicated. On the one hand, the new phase forsterite ((Mg,Fe)₂SiO₄) will be formed during the reduction process and the iron oxide in the phase is too hard to be reduced into metallic iron until the end of the reduction experiment. It can be proved by the next XRD analysis of the reduced pellets. On the other hand, at the later stage of reduction the contact between carbon and iron oxide is broken and CO partial pressure is very low, therefore, the reduction becomes more difficult.

Fig. 6.

Variation of metallization degree of coal char composite pellets over time at different C/O (1300°C).

3.2. Microscopic Examination for the Reduced Pellets

Figure 7 presents the cross-sectional SEM images (backscattered electrons) of the pellets reduced at 1300°C for different time. The chemical analysis can be carried out through the EDS with the SEM. When the pellet is reduced for 3 min, it is obvious that the macrostructure of the pellet has not changed much and is still the mixture of raw materials. When reduced for 6 min, some small metallic iron particles come into being. When reduced for 9 min, the metallic iron particles become more and bigger. For the 12 min reduced pellet, much more metallic iron was suddenly reduced out and the iron particles become much bigger. The size of iron particles and amount of metallic iron continue to increase as the reduction experiment continues. It can be seen that the microstructures of the pellets change little if the reduction time is longer than 15 min, which agrees well with the chemical analysis results. The diameter of the most of the iron particles are less than 50 μm and embedded around the gangue minerals, which are primarily forsterite, and residual carbon particles. The gangue minerals and carbon particles may act as the nucleation centers of the metallic iron particles.

Fig. 7.

Microstructure of the pellet during the reduction process. (a) 3 min, (b) 6 min, (c) 9 min, (d) 12 min, (e) 15 min, (f) 18 min.

3.3. Phase Evolution during the Reduction Process

The phase evolution during the reduction process is characterized by XRD analysis of the powder samples of the reduced pellets. The patterns of the reduced samples are shown in Fig. 8. The experiments are performed at 1300°C for different time at the interval of 3 min. The main phase in the original pellet is magnetite (Fe₃O₄). For 3 min, the intensity of magnetite peaks decreases sharply and it nearly disappears. The characteristic peaks of wustite (FeO), metallic iron (α-Fe), forsterite ((Mg,Fe)₂SiO₄) and ludwigite (Mg₃(Mg_0.4 Fe_0.6)Fe₂B₂O₁₀) come into being at the same time. It indicates that some of the iron oxide becomes more difficult to be reduced. For 6 min, the peaks of metallic iron increase and those of FeO decrease. When the pellet is reduced for 9 min, the wustite almost disappears and the intensity of metallic iron keeps on increasing. The peaks of metallic iron become strongest at the time of 12 min and it coexists with forsterite until the end of the reduction process.

Fig. 8.

XRD patterns of pellets during reduction process.

3.4. Kinetics Analysis of the Iron Oxide Reduction Process

In order to interpret the carbothermic reduction mechanism of boron-bearing iron concentrate, the rate controlling step and apparent activation energy should be determined. The reduction of iron oxide can be taken as the first order reaction.^13,15) The relation between reduction fraction and rate constant is expressed as follows,   

$$-ln(1-f)=kt$$(2)

Where f is overall reduction fraction of iron oxide, t is reduction time (s) and k is integrated rate constant (s^–1).

The f is calculated by Eq. (3).   

$$f=1-\frac{W_O^t}{W_O^0}$$(3)

Where $W_O^t$ is the weight of reducible oxygen in pellet, which is reduced for t min, $W_O^0$ is the weight of reducible oxygen in green pellet. All the two parameters are obtained through chemical analysis. The influence of volatile of coal and burning loss of iron ore on the accuracy of experiment data can be excluded in this method.

The reduction rate plots can be obtained when the ‘f’ and‘t’are introduced into Eq. (2), and are shown in Fig. 9. The gradient of the plotted line is taken as the rate constant k. The k of different temperatures abides by the Arrhenius Equation, i.e. Eq. (4).   

$$k=k_{0}exp\left(-\frac{E}{RT}\right)$$(4)

  

$$lnk=ln{k}_0-\frac{E}{RT}$$(5)

Where k₀ is frequency factor (s^–1), R is ideal gas constant (8.314×10^–3 kJ/(mol·K)), E is apparent activation energy (kJ/mol), T is temperature (K).

Fig. 9.

Rate analysis of the reduction process.

It can be seen that the –ln(1–f) vs time curves, such as k₂ and k₄ lines in Fig. 9, are not linear throughout the reduction process and the exact reason for this isn’t completely understood. At the initial stage of the reduction, the temperature of the pellet is lower than the setting temperature for the reduction experiment and the reduction rate is theoretically lower. During the reduction process, the new phase forsterite ((Mg,Fe)₂SiO₄), which is harder to be reduced than FeO, will be formed. Moreover, the decrease of the surface area of both wustite and coal char as the reduction progresses, the penetration of inert gas inside the pellet with the subsequent lowering of CO and CO₂ partial pressure, and the decrease on the heat flux needed for the Boudouard reaction as the pellet reaches the furnace temperature are all may be the reasons.¹⁶⁾

The Arrhenius plots of the rate constants of the pellets are shown in Fig. 10. The apparent activation energy of reduction process of boron-bearing concentrate/coal composite pellet is 114.32 kJ/mol. The value is less than those obtained by Fruehan¹³⁾ (293 to 334 kJ/mol) and Rao¹⁵⁾ (301 kJ/mol), but is comparable to those obtained by Seaton et al.¹⁶⁾ (159 kJ/mol) and Abraham and Ghost¹⁷⁾ (140 to 296 kJ/mol). The activation energies (Ea) of different chemical reactions during iron oxide reduction are given in Table 3.¹⁸⁾ According to the value of apparent activation energy in the present reduction process, the reduction process may be mixed control by the Boudouard reaction and the chemical reaction of FeO reduction by CO at the mean time.

Fig. 10.

Arrhenius plot of the rate constants for the reduction process.

Table 3. Activation energies of different reactions during the reduction process.

Reaction	Ea/kJ/mol
(1/4)Fe3O4+CO=(3/4)Fe+CO2 (T<570°C)	73.6
Fe3O4+CO=3FeO+CO2 (T>570°C)	73.6
FeO+CO=Fe+CO2	69.4
C+CO2=2CO	221.8

3.5. Properties of Separated Iron and Boron-rich Slag

The reduced pellet can’t melt separation clearly at 1500°C. However, the iron and slag separate much better when heated at 1550°C for 40 min. The liquidus temperature of the slag formed during the melting process of the reduced composite pellet without considering FeO is about 1596°C calculated by FACT-SAGE package. The ratio of the liquid phase in the slag is about 74% and 88% at 1500°C and 1550°C, respectively. More liquid phase will form with the increasing of furnace temperature and the melting separation will become easier. The result indicates that the amount of liquid phase at 1550°C is enough for the iron and slag melting separation under the experimental condition. The morphology of the separated iron and slag is shown in Fig. 11. The chemical compositions of the iron and slag are listed in Tables 4 and 5. The sulfur content in the iron is very low and it meets the requirement of steel making and casting. The distribution ratio of sulfur (wt%(S)/wt%[S]) between slag and iron is 41.2 and nearly 95% of the total sulfur in the raw materials enters into the slag phase under the experimental conditions. According to the mass balance calculation, 23.0% B₂O₃ in the pellet is reduced into pig iron, 63.6% B₂O₃ beneficiates into the boron-rich slag and the other volatilizes into gas phase. The standard Gibbs free energies of chemical reactions of B₂O₃ reduction are listed in Fig. 12.^19,20) It is obvious that B₂O₃ is difficult to reduce and increasing temperature can improve the reduction reactions. However, the reduction of B₂O₃ into B by solid carbon becomes easier if the B can dissolve into molten iron and the partial pressure of CO is low enough. The chemical composition of the pig iron shows that the present experimental conditions meet the above requirements to some extent. During the reduction process, a 5 L/min flow of Nitrogen is used as the purge gas and the newly formed gaseous product (i.e. B₂O₂ and CO) will be carried away quickly. So, a certain amount of boron element will not be enriched into the slag phase and the yield of boron will then decrease.

Fig. 11.

Morphology of the separation iron and slag.

Table 4. Chemical composition of pig iron (wt%).

C	Si	B	P	S
3.29	2.35	0.74	0.055	0.050

Table 5. Chemical composition of boron-rich slag (wt%).

TFe	MFe	MgO	Al2O3	CaO	B2O3	SiO2	S
1.55	1.49	56.2	2.38	1.80	10.8	21.3	2.06

Fig. 12.

Standard Gibbs free energies of boron oxide reduction reactions.

The X-ray diffraction result of the slag is given in Fig. 13 and it shows that the main mineral phases in the slow cooling boron-rich slag are olivine, kotoite, periclase and spinel. The amounts of periclase and spinel are small. Most of the boron oxide in the slag forms the crystalline phase, i.e. kotoite. The microstructure of the slag examined by SEM-EDS is shown in Fig. 14. It is clear that there are mainly two kinds of phases: dark grey and light grey, except for the small white particle clusters, which mainly contain sulfur. According to the micrograph, the amount of light grey phase is more than the dark grey one. Through comprehensive consideration of XRD and EDS analysis results, the dark grey phase is kotoite and the light grey one is olivine. The particle diameter of olivine is bigger than kotoite and their morphology is similarly lath-shaped. Due to the reduction of B₂O₃ into molten iron and into gaseous phase, the MgO component is isolated and forms the mineral phase of periclase during cooling crystallization process. The morphology and dissemination characteristics of the periclase are given in Fig. 15. It can be seen that most of the periclase particles are in round shape, the diameters of which are about 10–20 μm. The Sulfur element in the boron-rich slag embeds on the boundary of olivine and kotoite with calcium element. The EEB of the slag is only 68.4% and is much lower than our former work,¹²⁾ which is as high as 86.46%. The main reason for this may be that the reactivity of kotoite isn’t good enough and the B₂O₃ content in the slag is lower.

Fig. 13.

XRD analysis of the boron-rich slag.

Fig. 14.

SEM and EDS (mapping) analysis of the boron-rich slag.

Fig. 15.

SEM and EDS analysis of the periclase in boron-rich slag.

4. Conclusions

(1) The reduction of boron-bearing iron concentrate by coal proceeded at a visible rate when the temperature was higher than 1200°C due to the restriction of Boudouard reaction. The metallization degree of reduced pellet can reach more than 90% only when the heating temperature is higher than 1300°C.

(2) Carbon content (C/O ratio) has limited effect on the reduction rate of the pellet at the temperature of 1300°C. The C/O should be more than 0.8, however, it may be still enough if it is less than 1.2. The appropriate reduction time is around 18 min and the metallization degree will not increase if the time is prolonged.

(3) The metallic iron increases with the increasing of reduction time at 1300°C. The metallic iron particles increased dramatically when the pellet was reduced at 1300°C for more than 9 min. The metallic iron particles agglomerate and surround around the residual carbon and gangue minerals.

(4) The iron oxide of the composite pellet is reduced in the sequence of Fe₃O₄→FeO→Fe and Fe₃O₄ disappears quickly at the initial reduction step. The apparent activation energy of the reduction process is 114.32 kJ/mol. The reduction rate may be controlled simultaneously by Boudouard reaction and FeO reduction.

(5) The reduced pellet separates well at 1550°C, and nearly 63.6% B₂O₃ in the ore and 95% of the total sulfur in the raw materials beneficiate into the slag phase. After slow cooling, the main phases in the slag are olivine, kotoite, periclase and spinel, with olivine and kotoite dominating. The EEB of the slag is only 68.4% and its property should be further improved. The separation iron is good raw material for further utilization due to its low sulfur content and bearing boron.

Acknowledgments

The authors would like to express their gratitude for the financial support of the National Natural Science Foundation of China (Grant NO. 51274033 and 51374024) as well as the research grant from the State Key Laboratory of Advanced Metallurgy, University of Science and Technology Beijing.

References

• 1)   S.  Ken and  K. M.  Marcia: Mineral Commodity Summaries 2012, ed. by U.S. Geological Survey, USGS, Reston, VA, (2012), 33.
• 2)   X. P.  Zhang,  J. F.  Lang,  C. M.  Cui and  S. L.  Liu: Iron Steel, 30 (1995), 9.
• 3)   R.  Liu,  X. X.  Xue,  T.  Jiang,  S. H.  Zhang and  D. W.  Huang: Multi. Utili. Miner. Res., (2006), No. 2, 33.
• 4)   K. H.  Bauer,  D.  Huette,  H. J.  Lehmkuehler and  H.  Schmauch: Metall. Plant Technol., 13 (1990), 74.
• 5)   J. M.  Mcclelland and  G. E.  Metius: JOM, 55 (2003), 30.
• 6)   R.  Munnix,  J.  Borlee and  D.  Steyls: Metall. Plant Technol., 20 (1997), 50.
• 7)   J. C.  Simmons,  K. J.  Shoop and  J. M.  Mcclelland: Direct From Midrex, (2003), 3.
• 8)   Y.  Sawa,  T.  Yamamoto,  K.  Takeda and  H.  Itaya: ISIJ Int., 41 (2001), 17.
• 9)   B.  Anameric and  S. K.  Kawatra: Miner. Metall. Proc., 23 (2006), 52.
• 10)   S. L.  Liu,  C. M.  Cui and  X. P.  Zhang: ISIJ Int., 38 (1998), 1077.
• 11)   Q. J.  Zhao,  F. M.  Meng and  Z. B.  Ma: J. Northeast Univ. Technol., 12 (1991), 464.
• 12)   G.  Wang,  J. S.  Wang,  Y. G.  Ding,  S.  Ma and  Q. G.  Xue: ISIJ Int., 52 (2012), 45.
• 13)   R. J.  Fruehan: Metall. Trans. B, 8 (1977), 279.
• 14)   J.  Yang and  T.  Mori: ISIJ Int., 47 (2007), 1394.
• 15)   Y. K.  Rao: Metall. Trans. B, 2B (1971), 1439.
• 16)   C. E.  Seaton,  J. S.  Foster and  J.  Relasco: Trans. Iron Steel Inst. Jpn., 6 (1983), 490.
• 17)   M. C.  Abraham and  A.  Ghost: Ironmaking Steelmaking, 1 (1986), 14.
• 18)   K.  Sun and  W. K.  Lu: Metall. Mater. Trans. B, 40 (2009), 91.
• 19)   C. H.  Dai and  X. P.  Zhang: J. Northeastern Univ., 17 (1995), 381.
• 20)   Q. J.  Zhao: J. Northeastern Univ., 11 (1990), 122.